Skip to main content

Troubleshooting Aerospike Connect for Spark

Mistakes to avoid

Mixed data types in bins

In Aerospike, records with the same bin name can have mixed data types. Spark views each bin as a database column with the same data type. This can cause runtime errors in Spark.

To handle mixed data types, refer to aerospike.schema.flexible on the Aerospike Connect for Spark Configuration Reference page.

Very sparse bins

Aerospike Connect for Spark detects the schema by a sampling of records. This means the sampled records may not have some bins. To avoid this situation, provide the schema.

Common error messages

This section addresses error messages you may encounter when working with Aerospike Connect for Spark.

java.lang.IllegalArgumentException: Unsupported class file major version

Run java -version to verify that the Java version in use is 1.8.

Operation not allowed at this time

Error message: com.aerospike.client.AerospikeException: Error 22,1,30000,0,0,BB9030011AC4202 3000: Operation not allowed at this time

Check the aerospike.conf file. Aerospike versions 4.9 and later require nsup-period to be set explicitly in order to allow records with TTL to be inserted. Other root causes are described in Client Writes Fail With Aerospike Err Fail Forbidden.


Error message: java.util.concurrent.ExecutionException: com.aerospike.client.AerospikeException$Connection: Error -8,1,1000,1000,0,BB9040016AE4202 3000:

This happens when the server gets more scan requests than it can process. To avoid this problem, reduce the value of aerospike.partition.factor and configure scan-threads-limit on the server.


Error message: 22 AS_ERR_FORBIDDEN

Raise the scan-threads-limit dynamically and/or reduce the single-scan-threads so that no starvation is happening when a lot of parallel partition scans are kicked off.

You can also reduce the value of aerospike.partition.factor.